Implicit Solvent in Gromacs.

I have been unable to find one single and complete guide on implementing an implicit solvent in Gromacs ver 5+. So, after finally piecing together several archived group posts and fragments of mdp settings I have it working and here is how. However, before going ahead with this guide I would consider several issues:

1. The code is old. For example, the implicit solvent does not support the “verlet” neighbourhood search cut-off scheme, and rather requires the pre-4.6 version “group” option. Eventually, “group” will be dropped in future releases of Gromacs.
2. Given that periodic boundary conditions are meant to emulate the effects of a vast bulk solvent environment there is no longer any need for them. In addition, as there is currently no way of calculation long-range interactions in implicit solvents, features based on periodic boundary conditions (e.g., particle mesh Ewald) must be switched off as they do not work with the Generalised Born implicit solvent (GBIS). However, by switching them off consider:
3. Turning all cut-offs to infinity. Otherwise, in a non-periodic environment, the energy will not converge. However, this then makes running your code in parallel very difficult.

Here are the essential mdp options required for energy minimisation, NVT and NPT calculations (add additional options as appropriate).

cutoff_scheme = group
vdwtype            = cutoff
ns_type              =  grid
nstlist                 = 0
rlist                     = 0
rcoulomb          = 0
rvdw                  = 0
coulombtype    = cutoff
pbc                      = no
comm-mode      = angular
comm-grps        = Protein
implicit_solvent = GBSA
gb_algorithm      = OBC
nstgbradii            = 1
rgbradii                = 0
gb_epsilon_solvent  = 80
gb_saltconc                = 0
gb_obc_alpha            = 1
gb_obc_beta              = 0.8
gb_obc_gamma        = 4.85
gb_dielectric_offset = 0.009
sa_algorithm = ace_approximation

Considering the three points raised it won’t take long before you realise that your system does not scale. Removing explicit water molecules has, in fact, not helped improve the folding efficiency of Gromacs. Either stick with explicit water molecules in a reduced unit cell (for example, Tip3P) or as I’ve been told, run implicit solvent calculations on the Amber code.

Please Like or leave a comment if this has helped in any way.